/// <summary>
/// Read one RDDCommand from the stream.
/// </summary>
/// <param name="stream">Stream to read from</param>
/// <returns>RDDCommand object</returns>
private static RDDCommand ReadRDDCommand(Stream stream)
{
int commandBytesCount = SerDe.ReadInt32(stream);
if (commandBytesCount <= 0)
{
throw new InvalidDataException(
$"Invalid command size: {commandBytesCount}");
}
var rddCommand = new RDDCommand
{
WorkerFunction = new RDD.WorkerFunction(
CommandSerDe.Deserialize <RDD.WorkerFunction.ExecuteDelegate>(
stream,
out CommandSerDe.SerializedMode serializerMode,
out CommandSerDe.SerializedMode deserializerMode,
out var runMode))
};
rddCommand.SerializerMode = serializerMode;
rddCommand.DeserializerMode = deserializerMode;
return(rddCommand);
}
/// <summary>
/// Read SqlCommands from the stream.
/// </summary>
/// <param name="stream">Stream to read from</param>
/// <param name="evalType">Evaluation type for the current commands</param>
/// <returns>SqlCommand objects</returns>
private static SqlCommand[] ReadSqlCommands(
PythonEvalType evalType,
Stream stream)
{
int numUdfs = SerDe.ReadInt32(stream);
var commands = new SqlCommand[numUdfs];
for (int i = 0; i < numUdfs; ++i)
{
var command = new SqlCommand();
int numArgsOffsets = SerDe.ReadInt32(stream);
command.ArgOffsets = new int[numArgsOffsets];
for (int argIndex = 0; argIndex < numArgsOffsets; ++argIndex)
{
command.ArgOffsets[argIndex] = SerDe.ReadInt32(stream);
}
command.NumChainedFunctions = SerDe.ReadInt32(stream);
for (int funcIndex = 0; funcIndex < command.NumChainedFunctions; ++funcIndex)
{
int commandBytesCount = SerDe.ReadInt32(stream);
if (commandBytesCount > 0)
{
CommandSerDe.SerializedMode serializerMode;
CommandSerDe.SerializedMode deserializerMode;
if (evalType == PythonEvalType.SQL_SCALAR_PANDAS_UDF)
{
var curWorkerFunction = new ArrowWorkerFunction(
CommandSerDe.Deserialize <ArrowWorkerFunction.ExecuteDelegate>(
stream,
out serializerMode,
out deserializerMode,
out string runMode));
command.WorkerFunction = (command.WorkerFunction == null) ?
curWorkerFunction :
ArrowWorkerFunction.Chain(
(ArrowWorkerFunction)command.WorkerFunction,
curWorkerFunction);
}
else if (evalType == PythonEvalType.SQL_GROUPED_MAP_PANDAS_UDF)
{
if ((numUdfs != 1) || (command.WorkerFunction != null))
{
throw new InvalidDataException(
"Grouped map UDFs do not support combining multiple UDFs");
}
command.WorkerFunction = new ArrowGroupedMapWorkerFunction(
CommandSerDe.Deserialize <ArrowGroupedMapWorkerFunction.ExecuteDelegate>(
stream,
out serializerMode,
out deserializerMode,
out string runMode));
}
else
{
var curWorkerFunction = new PicklingWorkerFunction(
CommandSerDe.Deserialize <PicklingWorkerFunction.ExecuteDelegate>(
stream,
out serializerMode,
out deserializerMode,
out string runMode));
command.WorkerFunction = (command.WorkerFunction == null) ?
curWorkerFunction :
PicklingWorkerFunction.Chain(
(PicklingWorkerFunction)command.WorkerFunction,
curWorkerFunction);
}
command.SerializerMode = serializerMode;
command.DeserializerMode = deserializerMode;
}
else
{
throw new InvalidDataException(
$"Invalid command size: {commandBytesCount}");
}
}
commands[i] = command;
}
return(commands);
}