public void TestInvalidChainingArrow()
{
var func1 = new ArrowWorkerFunction(
new ArrowUdfWrapper <Int32Array, StringArray, StringArray>(
(numbers, strings) => (StringArray)ToArrowArray(
Enumerable.Range(0, strings.Length)
.Select(i => $"{strings.GetString(i)}:{numbers.Values[i]}")
.ToArray())).Execute);
var func2 = new ArrowWorkerFunction(
new ArrowUdfWrapper <StringArray, StringArray>(
(strings) => (StringArray)ToArrowArray(
Enumerable.Range(0, strings.Length)
.Select(i => $"outer1:{strings.GetString(i)}")
.ToArray())).Execute);
IArrowArray[] input = new[]
{
ToArrowArray(new[] { 100 }),
ToArrowArray(new[] { "name" })
};
// The order does not align since workerFunction2 is executed first.
ArrowWorkerFunction chainedFunc1 = ArrowWorkerFunction.Chain(func2, func1);
Assert.ThrowsAny <Exception>(() => chainedFunc1.Func(input, new[] { 0, 1 }));
}
public void TestChainingArrowWorkerFunction()
{
var func1 = new ArrowWorkerFunction(
new ArrowUdfWrapper <int, string, string>(
(number, str) => $"{str}:{number}").Execute);
var func2 = new ArrowWorkerFunction(
new ArrowUdfWrapper <string, string>(
(str) => $"outer1:{str}").Execute);
var func3 = new ArrowWorkerFunction(
new ArrowUdfWrapper <string, string>(
(str) => $"outer2:{str}").Execute);
Apache.Arrow.IArrowArray[] input = new[]
{
ToArrowArray(new[] { 100 }),
ToArrowArray(new[] { "name" })
};
// Validate one-level chaining.
var chainedFunc1 = ArrowWorkerFunction.Chain(func1, func2);
ArrowTestUtils.AssertEquals(
"outer1:name:100",
chainedFunc1.Func(0, input, new[] { 0, 1 }));
// Validate two-level chaining.
var chainedFunc2 = ArrowWorkerFunction.Chain(chainedFunc1, func3);
ArrowTestUtils.AssertEquals(
"outer2:outer1:name:100",
chainedFunc2.Func(0, input, new[] { 0, 1 }));
}
public void TestChainingArrowWorkerFunction()
{
var func1 = new ArrowWorkerFunction(
new ArrowUdfWrapper <Int32Array, StringArray, StringArray>(
(numbers, strings) => (StringArray)ToArrowArray(
Enumerable.Range(0, strings.Length)
.Select(i => $"{strings.GetString(i)}:{numbers.Values[i]}")
.ToArray())).Execute);
var func2 = new ArrowWorkerFunction(
new ArrowUdfWrapper <StringArray, StringArray>(
(strings) => (StringArray)ToArrowArray(
Enumerable.Range(0, strings.Length)
.Select(i => $"outer1:{strings.GetString(i)}")
.ToArray())).Execute);
var func3 = new ArrowWorkerFunction(
new ArrowUdfWrapper <StringArray, StringArray>(
(strings) => (StringArray)ToArrowArray(
Enumerable.Range(0, strings.Length)
.Select(i => $"outer2:{strings.GetString(i)}")
.ToArray())).Execute);
var input = new IArrowArray[]
{
ToArrowArray(new[] { 100 }),
ToArrowArray(new[] { "name" })
};
// Validate one-level chaining.
ArrowWorkerFunction chainedFunc1 = ArrowWorkerFunction.Chain(func1, func2);
AssertEquals(
"outer1:name:100",
chainedFunc1.Func(input, new[] { 0, 1 }));
// Validate two-level chaining.
ArrowWorkerFunction chainedFunc2 = ArrowWorkerFunction.Chain(chainedFunc1, func3);
AssertEquals(
"outer2:outer1:name:100",
chainedFunc2.Func(input, new[] { 0, 1 }));
}
public void TestInvalidChainingArrow()
{
var func1 = new ArrowWorkerFunction(
new ArrowUdfWrapper <int, string, string>(
(number, str) => $"{str}:{number}").Execute);
var func2 = new ArrowWorkerFunction(
new ArrowUdfWrapper <string, string>(
(str) => $"outer1:{str}").Execute);
Apache.Arrow.IArrowArray[] input = new[]
{
ToArrowArray(new[] { 100 }),
ToArrowArray(new[] { "name" })
};
// The order does not align since workerFunction2 is executed first.
var chainedFunc1 = ArrowWorkerFunction.Chain(func2, func1);
Assert.ThrowsAny <Exception>(() => chainedFunc1.Func(0, input, new[] { 0, 1 }));
}
/// <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);
}