Package org.apache.flink.table.runtime.dataview

Class DataViewUtils

java.lang.Object

org.apache.flink.table.runtime.dataview.DataViewUtils

@Internal
public final class DataViewUtils
extends Object

Utilities to deal with DataViews.

For aggregating functions: A DataView is a field that is either represented as a regular StructuredType or as a RawType that serializes to null when backed by a state backend. In the latter case, a DataViewSpec contains all information necessary to store and retrieve data from state.

For process table functions: A DataView is a top-level instance that is always backed by a state backend.

Nested Class Summary

Nested Classes

Modifier and Type	Class	Description
static class	DataViewUtils.DistinctViewSpec	Specification for a special MapView for deduplication.

Method Summary

Modifier and Type	Method	Description
static org.apache.flink.table.types.DataType	adjustDataViews(org.apache.flink.table.types.DataType accumulatorDataType, boolean hasStateBackedDataViews)	Modifies the data type of an accumulator regarding data views.
static void	checkForInvalidDataViews(org.apache.flink.table.types.logical.LogicalType type)	Checks that the given type and its children do not contain data views.
static org.apache.flink.table.types.DataType	createDistinctViewDataType(org.apache.flink.table.types.DataType keyDataType, int filterArgs, int filterArgsLimit)	Creates a special DataType for DISTINCT aggregates.
static DataViewUtils.DistinctViewSpec	createDistinctViewSpec(int index, org.apache.flink.table.types.DataType distinctViewDataType)	Creates a special DataViewUtils.DistinctViewSpec for DISTINCT aggregates.
static List<DataViewSpec>	extractDataViews(int aggIndex, org.apache.flink.table.types.DataType accumulatorDataType)	Searches for data views in the data type of an accumulator and extracts them.
static boolean	isDataView(org.apache.flink.table.types.logical.LogicalType viewType, Class<? extends org.apache.flink.table.api.dataview.DataView> viewClass)	Returns whether the given LogicalType qualifies as a DataView.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Method Detail

isDataView

public static boolean isDataView(org.apache.flink.table.types.logical.LogicalType viewType,
                                 Class<? extends org.apache.flink.table.api.dataview.DataView> viewClass)

Returns whether the given LogicalType qualifies as a DataView.

checkForInvalidDataViews

public static void checkForInvalidDataViews(org.apache.flink.table.types.logical.LogicalType type)

Checks that the given type and its children do not contain data views.

extractDataViews

public static List<DataViewSpec> extractDataViews(int aggIndex,
                                                  org.apache.flink.table.types.DataType accumulatorDataType)

Searches for data views in the data type of an accumulator and extracts them.

adjustDataViews

public static org.apache.flink.table.types.DataType adjustDataViews(org.apache.flink.table.types.DataType accumulatorDataType,
                                                                    boolean hasStateBackedDataViews)

Modifies the data type of an accumulator regarding data views.

For performance reasons, each data view is wrapped into a RAW type which gives it LazyBinaryFormat semantics and avoids multiple deserialization steps during access. Furthermore, a data view will not be serialized if a state backend is used (the serializer of the RAW type will be a NullSerializer in this case).

createDistinctViewDataType

public static org.apache.flink.table.types.DataType createDistinctViewDataType(org.apache.flink.table.types.DataType keyDataType,
                                                                               int filterArgs,
                                                                               int filterArgsLimit)

Creates a special DataType for DISTINCT aggregates.

createDistinctViewSpec

public static DataViewUtils.DistinctViewSpec createDistinctViewSpec(int index,
                                                                    org.apache.flink.table.types.DataType distinctViewDataType)

Creates a special DataViewUtils.DistinctViewSpec for DISTINCT aggregates.