Package org.apache.drill.exec.physical.resultSet.impl

Class TupleState.MapColumnState

java.lang.Object

org.apache.drill.exec.physical.resultSet.impl.ColumnState

org.apache.drill.exec.physical.resultSet.impl.ColumnState.BaseContainerColumnState

org.apache.drill.exec.physical.resultSet.impl.TupleState.MapColumnState

Enclosing class:

TupleState

public static class TupleState.MapColumnState
extends ColumnState.BaseContainerColumnState

Represents a map column (either single or repeated). Includes maps that are top-level, nested within other maps, or nested inside a union. Schema management is a bit complex:

Condition	Action
	Outside of Union	Inside of Union
Unprojected	N/A	Omitted from output
Added in prior batch	Included in output
Added in present batch, before overflow	Included in output
Added in present batch, after overflow	Omitted from output this batch (added next batch)	Included in output

The above rules say that, for maps in a union, the output schema is identical to the internal writer schema. But, for maps outside of union, the output schema is a subset of the internal schema with two types of omissions:

• Unprojected columns
• Columns added after overflow
</ul

New columns can be added at any time for data readers that discover their schema as data is read (such as JSON). In this case, new columns always appear at the end of the map (remember, in Drill, a "map" is actually a structured: an ordered, named list of columns.) When looking for newly added columns, they will always be at the end.

Nested Class Summary

Nested classes/interfaces inherited from class org.apache.drill.exec.physical.resultSet.impl.ColumnState

ColumnState.BaseContainerColumnState, ColumnState.PrimitiveColumnState, ColumnState.State

Field Summary

Fields

Modifier and Type	Field	Description
protected boolean	isVersioned
protected final TupleState.MapState	mapState
protected final ColumnMetadata	outputSchema

Fields inherited from class org.apache.drill.exec.physical.resultSet.impl.ColumnState

addVersion, cardinality, loader, outputIndex, state, vectorState, writer

Constructor Summary

Constructors

Constructor	Description
MapColumnState(TupleState.MapState mapState, AbstractObjectWriter writer, VectorState vectorState, boolean isVersioned)

Method Summary

Modifier and Type	Method	Description
ContainerState	container()
boolean	isProjected()
boolean	isVersioned()	Indicate if this map is versioned.
TupleState.MapState	mapState()
ColumnMetadata	outputSchema()

Methods inherited from class org.apache.drill.exec.physical.resultSet.impl.ColumnState.BaseContainerColumnState

close, harvestWithLookAhead, rollover, startBatch, updateCardinality

Methods inherited from class org.apache.drill.exec.physical.resultSet.impl.ColumnState

allocateVectors, buildOutput, dump, innerCardinality, outerCardinality, schema, vector, vectorState, writer

Methods inherited from class java.lang.Object

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

Field Details

mapState

protected final TupleState.MapState mapState

isVersioned

protected boolean isVersioned

outputSchema

protected final ColumnMetadata outputSchema

Constructor Details

MapColumnState

public MapColumnState(TupleState.MapState mapState,
 AbstractObjectWriter writer,
 VectorState vectorState,
 boolean isVersioned)

Method Details

mapState

public TupleState.MapState mapState()

container

public ContainerState container()

Specified by:

container in class ColumnState.BaseContainerColumnState

isProjected

public boolean isProjected()

Overrides:

isProjected in class ColumnState

isVersioned

public boolean isVersioned()

Indicate if this map is versioned. A versionable map has three attributes:

1. Columns can be unprojected. (Columns appear as writers for the client of the result set loader, but are not materialized and do not appear in the projected output container.
2. Columns appear in the output only if added before the overflow row.
3. As a result, the output schema is a subset of the internal input schema.