org.apache.flink.table.sources

Interface PartitionableTableSource

All Known Implementing Classes:

HiveTableSource

@Experimental
public interface PartitionableTableSource

An interface for partitionable TableSource.

A PartitionableTableSource can exclude partitions from reading, which includes skipping the metadata. This is especially useful when there are thousands of partitions in a table.

A partition is represented as a Map<String, String> which maps from partition field name to partition value. Since the map is NOT ordered, the correct order of partition fields should be obtained via partition keys of catalog table.

Method Summary

Modifier and Type	Method and Description
TableSource	applyPartitionPruning(List<Map<String,String>> remainingPartitions) Applies the remaining partitions to the table source.
List<Map<String,String>>	getPartitions() Returns all the partitions of this PartitionableTableSource.

Method Detail

getPartitions

List<Map<String,String>> getPartitions()

Returns all the partitions of this PartitionableTableSource.

applyPartitionPruning

TableSource applyPartitionPruning(List<Map<String,String>> remainingPartitions)

Applies the remaining partitions to the table source. The remainingPartitions is the remaining partitions of getPartitions() after partition pruning applied.

After trying to apply partition pruning, we should return a new TableSource instance which holds all pruned-partitions.

Parameters:

remainingPartitions - Remaining partitions after partition pruning applied.

Returns:

A new cloned instance of TableSource holds all pruned-partitions.