Package edu.caltech.nanodb.plannodes

Class NestedLoopJoinNode

java.lang.Object

edu.caltech.nanodb.plannodes.PlanNode

edu.caltech.nanodb.plannodes.ThetaJoinNode

edu.caltech.nanodb.plannodes.NestedLoopJoinNode

All Implemented Interfaces:

java.lang.Cloneable

public class NestedLoopJoinNode
extends ThetaJoinNode

This plan node implements a nested-loop join operation, which can support arbitrary join conditions but is also the slowest join implementation.

Field Summary

Fields

Modifier and Type	Field	Description
private boolean	done	Set to true when we have exhausted all tuples from our subplans.
private Tuple	leftTuple	Most recently retrieved tuple of the left relation.
private static org.apache.logging.log4j.Logger	logger	A logging object for reporting anything interesting that happens.
private Tuple	rightTuple	Most recently retrieved tuple of the right relation.

Fields inherited from class edu.caltech.nanodb.plannodes.ThetaJoinNode

joinType, leftSchema, leftStats, predicate, rightSchema, rightStats, schemaSwapped

Fields inherited from class edu.caltech.nanodb.plannodes.PlanNode

cost, environment, leftChild, rightChild, schema, stats

Constructor Summary

Constructors

Constructor	Description
NestedLoopJoinNode(PlanNode leftChild, PlanNode rightChild, JoinType joinType, Expression predicate)

Method Summary

Modifier and Type	Method	Description
private boolean	canJoinTuples()
void	cleanUp()	Perform any necessary clean up tasks.
protected PlanNode	clone()	Creates a copy of this plan node and its subtrees.
boolean	equals(java.lang.Object obj)	Checks if the argument is a plan node tree with the same structure, but not necessarily the same references.
Tuple	getNextTuple()	Returns the next joined tuple that satisfies the join condition.
private boolean	getTuplesToJoin()	This helper function implements the logic that sets leftTuple and rightTuple based on the nested-loops logic.
int	hashCode()	Computes the hash-code of the nested-loop plan node.
void	initialize()	Do initialization for the join operation.
void	markCurrentPosition()	Marks the current tuple in the tuple-stream produced by this node.
void	prepare()	This method is responsible for computing critical details about the plan node, such as the schema of the results that are produced, the estimated cost of evaluating the plan node (and its children), and statistics describing the results produced by the plan node.
boolean	requiresLeftMarking()	True if the node requires that its left child supports marking.
boolean	requiresRightMarking()	True if the node requires that its right child supports marking.
void	resetToLastMark()	Resets the node's tuple-stream to the most recently marked position.
java.util.List<OrderByExpression>	resultsOrderedBy()	Nested-loop joins can conceivably produce sorted results in situations where the outer relation is ordered, but we will keep it simple and just report that the results are not ordered.
boolean	supportsMarking()	True if the node supports position marking.
java.lang.String	toString()	Returns a string representing this nested-loop join's vital information.

Methods inherited from class edu.caltech.nanodb.plannodes.ThetaJoinNode

isSwapped, joinTuples, prepareSchemaStats, swap

Methods inherited from class edu.caltech.nanodb.plannodes.PlanNode

addParentEnvironmentToPlanTree, duplicate, getCost, getEnvironment, getSchema, getStats, printNodeTree, printNodeTree, printNodeTree, printNodeTreeToString, printNodeTreeToString, setEnvironment

Methods inherited from class java.lang.Object

finalize, getClass, notify, notifyAll, wait, wait, wait

Field Detail

logger

private static org.apache.logging.log4j.Logger logger

A logging object for reporting anything interesting that happens.

leftTuple

private Tuple leftTuple

Most recently retrieved tuple of the left relation.

rightTuple

private Tuple rightTuple

Most recently retrieved tuple of the right relation.

done

private boolean done

Set to true when we have exhausted all tuples from our subplans.

Constructor Detail

NestedLoopJoinNode

public NestedLoopJoinNode(PlanNode leftChild,
                          PlanNode rightChild,
                          JoinType joinType,
                          Expression predicate)

Method Detail

equals

public boolean equals(java.lang.Object obj)

Checks if the argument is a plan node tree with the same structure, but not necessarily the same references.

Specified by:

equals in class PlanNode

Parameters:

obj - the object to which we are comparing

hashCode

public int hashCode()

Computes the hash-code of the nested-loop plan node.

Specified by:

hashCode in class PlanNode

Returns:

the hash code for the plan node and any subnodes it may contain.

toString

public java.lang.String toString()

Returns a string representing this nested-loop join's vital information.

Specified by:

toString in class PlanNode

Returns:

a string representing this plan-node.

clone

protected PlanNode clone()
                  throws java.lang.CloneNotSupportedException

Creates a copy of this plan node and its subtrees.

Overrides:

clone in class PlanNode

Throws:

java.lang.CloneNotSupportedException

resultsOrderedBy

public java.util.List<OrderByExpression> resultsOrderedBy()

Nested-loop joins can conceivably produce sorted results in situations where the outer relation is ordered, but we will keep it simple and just report that the results are not ordered.

Specified by:

resultsOrderedBy in class PlanNode

Returns:

If the plan node produces ordered results, this will be a list of objects specifying the ordering. If the node doesn't produce ordered results then the return-value will either be an empty list or it will be null.

supportsMarking

public boolean supportsMarking()

True if the node supports position marking.

Overrides:

supportsMarking in class PlanNode

Returns:

true if the node supports position marking, false otherwise.

requiresLeftMarking

public boolean requiresLeftMarking()

True if the node requires that its left child supports marking.

Overrides:

requiresLeftMarking in class PlanNode

Returns:

true if the node requires that its left child supports marking, false otherwise.

requiresRightMarking

public boolean requiresRightMarking()

True if the node requires that its right child supports marking.

Overrides:

requiresRightMarking in class PlanNode

Returns:

true if the node requires that its right child supports marking, false otherwise.

prepare

public void prepare()

Description copied from class: PlanNode

This method is responsible for computing critical details about the plan node, such as the schema of the results that are produced, the estimated cost of evaluating the plan node (and its children), and statistics describing the results produced by the plan node.

Specified by:

prepare in class PlanNode

initialize

public void initialize()

Description copied from class: ThetaJoinNode

Do initialization for the join operation. Resets state variables. Initialize both children.

Overrides:

initialize in class ThetaJoinNode

getNextTuple

public Tuple getNextTuple()

Returns the next joined tuple that satisfies the join condition.

Specified by:

getNextTuple in class PlanNode

Returns:

the next joined tuple that satisfies the join condition.

getTuplesToJoin

private boolean getTuplesToJoin()

This helper function implements the logic that sets leftTuple and rightTuple based on the nested-loops logic.

Returns:

true if another pair of tuples was found to join, or false if no more pairs of tuples are available to join.

canJoinTuples

private boolean canJoinTuples()

markCurrentPosition

public void markCurrentPosition()

Description copied from class: PlanNode

Marks the current tuple in the tuple-stream produced by this node. The PlanNode.resetToLastMark() method can be used to return to this tuple. Note that only one marker can be set in the tuple-stream at a time.

Overrides:

markCurrentPosition in class PlanNode

resetToLastMark

public void resetToLastMark()
                     throws java.lang.IllegalStateException

Description copied from class: PlanNode

Resets the node's tuple-stream to the most recently marked position. Note that only one marker can be set in the tuple-stream at a time.

Overrides:

resetToLastMark in class PlanNode

Throws:

java.lang.IllegalStateException - if PlanNode.markCurrentPosition() hasn't yet been called on this plan-node

cleanUp

public void cleanUp()

Description copied from class: PlanNode

Perform any necessary clean up tasks. This should probably be called when we are done with this plan node.

Specified by:

cleanUp in class PlanNode